Where are the guns?

We test the effectiveness of several common gun prevalence proxy variables against what are arguably the best measures of gun prevalence: firearm sales and concealed carry permits. With a comprehensive count of gun sales and concealed carry permits (by county and year) in the states of Pennsylvania and Massachusetts, we make two main observations: First, gun sales/concealed carry permits are positively correlated with federal firearm licenses (gun dealers) per mile. Second, gun sales/concealed carry permits are not significantly positively correlated with federal firearm licenses per capita or the proportion of gun suicides to total suicides. We then discuss why this occurs and the limitations of using legal gun sales as a gun prevalence measure. Last, we show how the competing measures differ in terms of their associations with gun homicide. We find our preferred measure to have a strong positive association with gun homicides while many others do not. Consequently, we advise researchers to use gun dealers as a measure of gun prevalence and specifically in a way that considers markets bleeding over arbitrary lines (e.g., county, city, or neighborhood). This will especially be the case if one is interested in small geographic areas.